@UnrelatedString

hostile takeover time

ah there

so what do you have so far

I probably.. want to use the range ascending thing?

what for

Idk this is my first brachylog thing I have no cluse

lol

so first off uh

@UnrelatedString to filter out non alphabets

I just thought it would be fitting to do it in brachylog

i was about to say your solution should probably be something all inside {...}ˢ but using a generator with ∋ is probably shorter

let me get a test harness for you

...wish there was a better set of canonical test cases

i'll just steal the jelly one

s

lool

okay so if you want to do a generator use this, if you want to do not a generator just remove the header

so yeah first thing for working towards a solution is that range fails if you try to give it something that isn't integers, but you can also use the built in alphabet string constants

does brachylog have builtin constants?

github.com/JCumin/Brachylog/wiki/8.3)-Variables-and-Constants

interesting

so ⊆Ạ should imply that the string is a subset of the alphabet

right

i'm not even sure that using Ạ would be the shortest way to remove words containing non-alphabetic characters but yeah

man I wish I was there during the brachylog lotm

the issue with using ⊆ here is that it's not actually subset, it's subsequence

so ⊆Ạ is "input consists entirely of lowercase letters of the alphabet, which are in strictly increasing order"

huh.

which sounds perfect except the "strictly" part

strictly increasing is..right?

"aa" would fail

oh.

come to think of it the actual challenge spec isn't completely clear about it but the remark about "Maddox" being "abecedarian in ASCII` implies "aa" should succeed

so you could use d to remove duplicates, but that would also remove duplicates which would otherwise be out of order

harsh

ḅhᵐ is one option, take the first element of every run

...maybe with a c on the end

:/

so that's coming out to 10 bytes

wait lemme check what those symbols do

yep

h is the first of each

h is just the first, then ᵐ maps it

oh, there's mapping

yeah

the only thing brachylog has that resembles implicit vectorization is backtracking so you're going to be seeing that a lot

backtracking hmm

so ḅhᵐ⊆Ạ?

yep

huh.

except you also need an ḷ on the start to casefold it

it gives an empty list

therefore the test case is wrong :P

and then you have to deal with it not actually performing any kind of filter

wrap it in braces

if you removed the header and then ran it on each element of the test case, it would succeed for the abecedarian words and fail for the other ones

neat.

I shall scour answers for inspiration

yeah here's a version that uses braces to filter

it's one byte longer than the generator solution

on a side note, i found another 10-byter that contains the substring "LoL"

the mutilated ryze jpeg gives you unforeseen power

one might say, desperate power

that is what his old ult was called right

i thought it was explode or something

that may have been an older one

how many reworks have they done to the poor man

i think the one from four years ago was his first proper vgu but i think he's had at least one mini rework since then and like four before

@UnrelatedString this is correct

original ryze ultimate was called "Desperate power"

makes everything he has into an AoE

i think it was just a ton of cdr and spell vamp at the end but yeah that sounds about right

ok back to trying (and failing) to find the 10

yeah

the select metapredicate now makes sense

so it's a shame you have to not use it for 10 lol

but yeah gj figuring that out faster than i did

lmao

i think i was writing brachylog for at least a week before i learned the distinction between ᵐ and ˢ

however, i don't get what the & is there for

can't find it in the docs

i think it's under control

@UnrelatedString I have been looking at the docs over and over and then not writing brachylog for a while now

but it's pretty much just an abbreviation of ∧?

as for why it's there, i could tell you but it's funnier to tell you to remove it

@UnrelatedString "impllicit unification" means that the slphabet just replaces the thing?

this is so wack

ah so & translates to ∧?

if the thing is false i suppose it short circuits

yeah, ˢ isn't so much a filter metapredicate as it is an analogue of ᵐ that handles failures differently

m just results in false

exactly

it would succeed only if the predicate didn't fail for any element

so creating a boolean array has to be done some other way

since ˢ is just ᵐ except it ignores elements that fail, to return the elements being filtered rather than some intermediate value in determining if they should be filtered out, you have to explicitly unify the input and output

...not entirely sure what you mean by boolean array

there is actually no concept of a truthy or falsy value, just success or failure

well err ok

@UnrelatedString why explicitly unify when implicit unification is done?

well actually implicit unification of adjacent variables is the only way things can be unified

and then input and output are implicitly present at the start and end of a predicate respectively

but what i mean by explicitly unify is that you have to place the input there explicitly to unify it with the output because it won't be unified with the output otherwise

so you just put a ? at the end

yep

and then you end up needing ∧ as well because {...Ạ?} would also try to unify the input with the alphabet

ah logical conjunction with the input, which is unified with the output

My brain is not fried enough yet

i think the conceptual reason ∧ is "just break unification lol" is that it represents the conjunction of two separate chains of unified stuff, as opposed to ∨ which breaks unification but creates a disjunction instead

interesting

i wonder how fast prolog is

i think swipl is pretty well optimized

i wonder how well it does in fastest-code

i think it wouldn't hurt to have meta predicates implicitly close curly braces

although it kinda messes with the aesthetic

i think that would be hard to disambiguate when you're also trying to use metapredicates on non compound predicates

I think brachylog would greatly benifit froma trace command if it doesn't already have one

@UnrelatedString I still need to understand that then

what kind of trace

tracing througha predicate and seeing what happens

showing what's getting unified and what result is coming out, so on

i just stick a bunch of {w&}s in the middle

which isn't always the best because it labels unbound variables

but uuuuuusually does the job

ah the joys of golfing

also you could use w₂ but you can't type that as fast lmao

I still need to get the brachlylog lb

doesn't the lb have keyboard shortcuts?

the bookmarklet? yeah

i never remember most of them lmao

well the subscripts are just underscore

I remember stuff only for apl

husk i just resort to intense clicking

yep

if they simply used the haskell names I would've like it a lot more

the husk and jelly bookmarklets also have the advantage of having documentation built in

maybe I should pr

@UnrelatedString I can add docs

ooh